Tglad's Formula crash error v1.21

[geomagnet]

OpenCL Build log:   :1733:1: error: expected expression
     float tgladDE = 1.0f;
^
:1759:3: error: use of undeclared identifier 'tgladDE'
                tgladDE = tgladDE * fabs(m) + 1.0f;
                ^
:1770:20: error: use of undeclared identifier 'tgladDE'
                        dist = r / fabs(tgladDE);

Hello,
I got this error trying to render Tglad's formula in OpenCL mode.  I'm sure it's looking for the declaration in the init file but I can't find which one it is.
How can I fix this?   

Thanks.

PS...great job on the latest upgrades.  the png alpha channel in OpenCL is very exciting.
  Allows for some interesting compositions.

[Pauldelbrot]

The declaration is there, but the first error shows it's not parsing it for some reason. If the actual file has a tab character at the start of that line, try changing it to spaces and see if that helps. It may be that whatever reads that file is allergic to tabs. If that doesn't do it then it's probably a problem that requires more context (more of the lines around the declaration) to figure out, and/or specific knowledge of OpenCL that I don't possess.

[mclarekin]

Hmmm, I don't have that problem ??.  Its declared in   cl_mandelbox_fullInit at the beginning

Code:

	float tgladDE = 1.0f;
	float3 ones = 1.0f;
	float3 foldingLimit = consts->fractal.mandelbox.foldingLimit;
	float3 foldingValue = consts->fractal.mandelbox.foldingValue;
	float mr2 = consts->fractal.mandelbox.minRadius * consts->fractal.mandelbox.minRadius;
	float fr2 = consts->fractal.mandelbox.fixedRadius * consts->fractal.mandelbox.fixedRadius;
	float scale = consts->fractal.mandelbox.scale;
	colourMin = 0.0f;

[quaz0r]

i just tried tglad in opencl mode with 1.21-1 on linux/nvidia and it completed a render without error.  however i just looked and cl_mandelbox_fullInit.cl does have actual tabs at the beginning of each line.  it would be typical Fisher-Price-OS fashion to fail on something like a tab.

[geomagnet]

Okay, so you're saying that if I open the "Fractal" tab and select "Tglad's formula (Mandelbox)" from the drop down and then
open the "Open CL" tab and Enable "OpenCL Kernel" and render, it uses the "cl_mandelbox_full*" files.   

I'm using notepad++.  I've tried ANSII and UTF-8.  I've tried removing the tab and replacing with space.  It doesn't seem to be
happy with anything I do.   Am I looking in the right place?   C://Program Files/Mandelbulber/cl/


Thanks
James

[mclarekin]

Either the "cl_mandelbox_ full" or the "cl_mandelbox", I am not sure if the mandelbox_full is only initiated when the "Enable Rotation" is ticked.

 C://Program Files/Mandelbulber/cl/ is the correct location.

 However from my little knowledge I suspect there is nothing really wrong with the code as it seems to work for most people, but with minor glitches that are more individual computer specific (ie driver software, GPU etc)

I guess you have tried reinstalling the programme and/or earlier versions to see if there is you get the same error?

[geomagnet]

I haven't gone to an older version, but as I recall a majority of the formulas didn't work with OpenCL mode on my system so I either dealt with the long rendering
times of the fractal I wanted, or found fast rendering fractals the work in OpenCL. Nothing new, I was hoping to be able to use OpenCL on all the fractals.
I guess the biggest problem with the error is that previous versions would crash but leave the program running in normal CPU mode.  Now it reports the error and
then exits mandelbulber completely (not giving you the chance to save any settings).

FYI - I'm running

Win 7 Ultimate 64-bit
on Intel core-i7 3770 with Intel motherboard DZ77-BH55
and Nvidia GTX650Ti
Not sure if that helps.

Thanks anyhow, we'll worry about it later.  I'll just stay away from the things that don't work.